On Sun, Jun 17, 2001 at 06:40:52PM -0700, Ian Clarke wrote:
> On Mon, Jun 18, 2001 at 03:30:58AM +0200, Oskar Sandberg wrote:
> > Firstly, I believe this is a bed pissing solution (nice and warm at
> > first, cold and sticky soon thereafter). Greater HTL values means more
> > caching, not less, and thus if that truly is the problem then increasing
> > the HTL will only make it worse. 
> 
> The worst caching is that closest to the requestor, a higher HTL makes
> this less significant in proportion to caching on nodes which are likely
> to be closer to the data epicenter - so your assertion is not
> nescessarily true.

I think if you need HTL values around 100 on a comparetively small
network then there is no epicenter topology what so ever. Theo noted a
while ago that small world networks allow for pretty good search
performance without any ordering at all - I think that is the only
positive effect you are seeing at the moment.

> > The argument that high HTL values
> > flaten the search trees also seems reasonable.
> 
> You say this like it is a bad thing.  It is possible that due to the
> seemingly rapid rate at which new nodes are being added to the network,
> that it is getting "strung out", and insufficient path-compression is
> taking place to address this problem.  A higher maximum HTL could help
> to alleviate this problem.

It certainly can be bad thing - too flat trees is just as bad as too
steep ones. What is actually going on at the moment is just pure
speculation.

-- 
'DeCSS would be fine. Where is it?'
'Here,' Montag touched his head.
'Ah,' Granger smiled and nodded.

Oskar Sandberg
oskar at freenetproject.org

_______________________________________________
Devl mailing list
Devl at freenetproject.org
http://lists.freenetproject.org/mailman/listinfo/devl

Reply via email to